• xref CI: failing numpy_dev build on 1.2.x #39205 can close this now probably

restarted https://github.com/pandas-dev/pandas/runs/1716733247 to confirm

1.2.x is green, but we have different tests failing here now. Will investigate

Could someone restart numpy dev ci? The failing tests were not touched by this revert and can not reproduce this locally.

failure is always with calls like:

df = pd.DataFrame({"A": [1, 2, 3], "B": [4, 5, 6]})
df.A

The second line is raising AttributeError: 'DataFrame' object has no attribute 'A' in this ci run

The second line is raising AttributeError: 'DataFrame' object has no attribute 'A' in this ci run

see that occasionally, first seen around 3rd Jan #38852 (comment)

This seems to have spread compared to the link you've posted. We got a lot of failures in test_merge too. Was someone able to debug this?

This seems to have spread compared to the link you've posted. We got a lot of failures in test_merge too. Was someone able to debug this?

IIRC locally i had these failures consistently but after the next merge of master resolved. On the couple of occasions i've seen on ci, restarting ci resolved. we'll see what happens here.

Seems to have worked here too.

I think OK to merge this. the other test failure is also a flaky test which will pass after one or more restarts eg. restarted on #39139 and passed on second attempt.

=========================== short test summary info ===========================
FAILED pandas/tests/io/pytables/test_store.py::test_coordinates - AssertionEr...
= 1 failed, 98979 passed, 4748 skipped, 1034 xfailed, 3 xpassed, 82 warnings in 1184.42s (0:19:44) =

Thanks @phofl

Yeah this is an ugly one, I set up a development env on windows a few days ago but could not reproduce this.